Resource Selection by Southeastern Fox Squirrels in a Fire-Maintained Forest System

Journal of Mammalogy, 97(2):631–638, 2016 DOI:10.1093/jmammal/gyv210 Published online January 11, 2016

Resource selection by southeastern fox squirrels in a fire-maintained forest system

Annemarie Prince,* M. Colter Chitwood, Marcus A. Lashley, Christopher S. DePerno, and Christopher E. Moorman Washington Department of Fish and Wildlife, P.O. Box 350, Colville, WA 99114, USA (AP) North Carolina State University, Department of Forestry and Environmental Resources, Box 7646, Raleigh, NC 27695, USA (MCC, MAL, CSD, CEM) * Correspondent: [email protected] Downloaded from Fire is essential to maintain the open forest structure required by the southeastern fox squirrel (Sciurus niger niger). In recent decades, managers of the longleaf pine (Pinus palustris) ecosystem have transitioned from dormant-season to growing-season burns, which more effectively limit midstory hardwood encroachment. Similarly, aggressive hardwood removal programs have been employed to further reduce hardwood midstory.

However, fox squirrels are dependent on oaks (Quercus spp.) for food and cover; thus, it is unclear how growing- http://jmammal.oxfordjournals.org/ season burns and hardwood removal may affect habitat quality for fox squirrels. We used compositional analysis to investigate selection of home ranges within the study area by 48 radiocollared fox squirrels on the Fort Bragg Military Installation, North Carolina. We used resource utilization functions with growing-season fire history and other habitat covariates as explanatory variables to test whether growing-season fires influenced the selection of habitat components within home ranges. Lastly, using a sample of fox squirrel relocations and paired random points, we performed binomial logistic regression to test whether habitat selection by fox squirrels was influenced by the availability of oaks and longleaf pines and select forest stand structural characteristics. When establishing home ranges, fox squirrels selected southern yellow pine over other cover types. Within home ranges, fox squirrel use increased with decreasing distance to a riparian area but was not affected by the application of growing-season by James Harper on April 21, 2016 fires. At the population level, fox squirrels selected for greater densities of reproductively mature oak stems. Fox squirrels likely benefit from growing-season fires that maintain expansive upland pine stands but are negatively affected by homogeneous fire application and mechanical hardwood removal that reduce the occurrence of reproductively mature oaks across the landscape. Managers should strive to maintain oaks in riparian areas, fire shadows, and naturally occurring patches within pine stands when managing for fox squirrels.

Anthropogenic fire has a long history in the southern United plant communities of the Southeast, fire also played a central States (Hudson 1982; Pyne 1982). Since their arrival over role in shaping the animal communities of the region. Indeed, 10,000 years ago, Native Americans burned southern pine many wildlife species are dependent on the vegetative struc- forests and grasslands to drive game and clear agricultural ture and resources maintained by frequent fires (Brennan et al. lands (Hudson 1982; Pyne 1982; Buckner 1989; MacCleery 1998). However, fire protection policies implemented in the 1993). This practice was continued by European immigrants early 20th century, coupled with land use changes, facilitated who burned the land for many of the same reasons (Stoddard the decline of fire-dependent plant communities and many of 1962; Pyne 1982). Before the arrival of European immigrants, the animal species they supported (Brennan 1991; Engstrom it is thought that most of the longleaf pine (Pinus palustris) et al. 1996; Brockway and Lewis 1997). forests of the Southeast burned every 2–10 years (Christensen Fire-dependent longleaf pine forests are important for south- 1981; Frost 1998), and after their arrival possibly every eastern fox squirrels (Sciurus niger niger; hereafter, fox squir- 1–3 years (Landers et al. 1990). Globally, longleaf pine com- rel), a species of high conservation priority in the region (Weigl munities have one of the most frequent fire return intervals et al. 1989; Perkins et al. 2008). The fox squirrel’s large body (Christensen 1981). Because of the influence fire had on the size is thought to be an adaptation for living in fire-maintained

631 632 JOURNAL OF MAMMALOGY pine forests, affording them increased mobility, access to by fox squirrels, but these burns can reduce the prevalence of widely spaced food resources, and the ability to manipulate mature hardwoods within longleaf pine forests (Robbins and large longleaf pine cones (Steele 1988; Weigl et al. 1989; Steele Myers 1992). Because fox squirrels rely heavily on acorns and and Weigl 1993). Declines in fox squirrel populations have other hard mast for a large percentage of their diet, the negative coincided with the degradation and loss of mature longleaf effects of fire on oaks and the subsequent decreased availability pine forests, which now occupy less than 3% of their original of hard mast could be limiting their populations (Baumgartner extent (Weigl et al. 1989; Kantola and Humphrey 1990; Loeb 1940; Allen 1943; Weigl et al. 1989; Kantola and Humphrey and Moncrief 1993; Landers et al. 1995; Perkins and Conner 1990; Greenberg and Simons 1999). Conversely, in the absence 2004). The drastic reduction of longleaf pine forests is credited of frequent fires, longleaf pine communities shift from open- to widespread timber harvest occurring at the turn of the 20th canopy forests to closed-canopy systems dominated by shade- century, rapid urbanization of the eastern United States, conver- tolerant and fire-sensitive plant species (Heyward 1939; Garren sion to slash (P. elliottii) or loblolly (P. taeda) pine plantations, 1943; Nowacki and Abrams 2008), a condition more suitable and fire suppression (Frost 1993; Landers et al. 1995; Outcalt for the eastern gray squirrel (S. carolinensis—Whitaker and and Sheffield 1996). Hamilton 1998).

The historical longleaf pine ecosystem was characterized by Currently, there is limited information on the effects of pre- Downloaded from widely spaced longleaf pine trees, scattered hardwood patches, scribed burning or hardwood removal on fox squirrels. Although and diverse understory vegetation (Frost 2006). Large, mature prescribed burning is commonly recommended for managing pine and hardwood trees provide important seasonal food fox squirrel habitat, these recommendations often do not specify resources for fox squirrels (Moore 1957; Ha 1983; Kantola a season or frequency for prescribed fire application (Weigl et al.

1986; Weigl et al. 1989). Additionally, large hardwood trees 1989; Conner et al. 1999; Conner and Godbois 2003; Perkins and http://jmammal.oxfordjournals.org/ serve as refugia and provide cavities for rearing young (Moore Conner 2004). Our objective was to investigate habitat selection 1957; Weigl et al. 1989, Kantola 1992; Conner and Godbois by southeastern fox squirrels at multiple ecological scales in an 2003). However, compared to longleaf pine, hardwood species area with a large-scale, growing-season fire regime and targeted are generally less tolerant of fire (though tolerance varies con- removal of oaks and other upland hardwoods. We predicted that siderably among hardwood species). Consequently, the extent fox squirrels would select upland pine stands but would con- of hardwoods within longleaf pine forests is limited by frequent centrate use in areas with remnant hardwoods, which should be and homogeneous prescribed fires (Lashley et al. 2014), and more prevalent in units with lower burn frequencies and in fire hardwoods often naturally occur on fire-maintained proper- shadows (e.g., moist soil depressions and drainages). ties only as individual canopy trees or in small isolated patches within the pine matrix (Greenberg and Simons 1999). by James Harper on April 21, 2016 Within the historical range of the longleaf pine ecosys- Materials and Methods tem, the focus of contemporary restoration and management Study area.—Fort Bragg Military Installation (hereafter, Fort practices is reduction of hardwood species that have invaded Bragg) is a 64,280-ha active military base in the Sandhills phys- pine uplands as a result of fire suppression (Provencher et al. iographic region of North Carolina, United States. Dominated 2001; Kush et al. 2004; Varner et al. 2005). In many cases, by an overstory of longleaf pine and an understory of wiregrass land managers use machinery, herbicides, and growing-season (Aristida spp.), Fort Bragg and other adjacent areas form the fire to achieve and maintain hardwood-free upland pine for- largest contiguous tract of longleaf pine-wiregrass ecosystem ests (Boyer 1990; Means 1996; Provencher et al. 2001; Varner remaining in North Carolina (Sorrie et al. 2006). Large hard- et al. 2005). However, a pure pine forest is not representative of wood trees, including turkey oak (Quercus laevis), sand post oak presettlement conditions (Frost 1993), and there is a growing (Q. stellata), blackjack oak (Q. marilandica), southern red oak concern about the negative ecological effects of oak (Quercus (Q. falcata), and hickory (Carya spp.), are scattered throughout spp.) reduction or removal on mast-dependent wildlife species the base and are present in small patches in the uplands, along like the fox squirrel (Hiers et al. 2014; Lashley et al. 2014). riparian areas and firebreaks (Lashley et al. 2014), and border- Historically, variation in fire regime and intensity allowed large ing parachute drop zones. Fort Bragg’s land managers use pre- canopy hardwood trees and isolated patches of smaller hard- scribed fire extensively to maintain an open forest midstory for wood trees to persist at 10–60 trees per hectare within pine- the federally endangered red-cockaded woodpecker (Picoides hardwood forests (Moore 1957; Frost 1993; Rebertus et al. borealis—Lashley et al. 2014). Beginning in 1989, prescribed 1993; Greenberg and Simons 1999). fires were conducted primarily during the growing season Although resource managers currently use dormant-season (April–June) every 3 years to prevent hardwood encroachment and growing-season prescribed fire to restore and maintain in the uplands (Lashley et al. 2014); however, dormant-season longleaf pine forests, some prescribed fire programs within the burns were conducted yearly in the parachute drop zones and southeastern United States are beginning to emphasize the tim- in areas not burned due to weather or lack of personnel the pre- ing of natural fires (i.e., lightning-ignited) and are shifting to vious year. Hunters at Fort Bragg were allowed to harvest 1 the use of more early growing-season prescribed fire (Cantrell fox squirrel per day with a season limit of 10 from October et al. 1995; Fill et al. 2012). Frequent prescribed fires during the to December. According to Fort Bragg harvest records, squir- growing season maintain the open forest conditions required rel hunter effort decreased since 1982, but fox squirrel harvest PRINCE ET AL.—SOUTHEASTERN FOX SQUIRREL RESOURCE SELECTION 633 remained relatively constant with an increasing trend since subset of relocations (n = 20) for each fox squirrel and returned 2008; on average, hunters harvested 78 fox squirrels annually to each relocation point. We set the relocation point as the plot from 2001 to 2011. center and used a fixed-radius plot (area = 0.04 ha—James and Animal capture and monitoring.—We trapped fox squirrels Shugart 1970) to measure diameter at breast height (DBH) for using wooden box traps (Baumgartner 1940) and wire cage all trees within the plot. For each tree species, we counted all traps (Model 103, Tomahawk Live Trap Company, Hazelhurst, trees with DBH ≥ 10 cm. For each fox squirrel relocation, we Wisconsin) baited with dried whole kernel corn. Trap locations repeated the protocol at a random point, which we established were chosen based on captures of fox squirrels in traps placed via a random bearing (0–360°) and distance (25–75 m away). randomly by Scott (2011). Once captured, we transferred fox Data analysis.—We assessed habitat selection of home squirrels into a modified capture cone (Koprowski 2002). Fox ranges using compositional analysis (Aebischer et al. 1993) squirrels were weighed, sexed, aged (juvenile or adult—Weigl as modified by Millspaugh et al. (2006); we compared cover et al. 1989), assessed for reproductive condition, and individu- types available within the study area to a utilization distribution ally ear-tagged (Monel 1005-1/1005-3, National Band and Tag (UD)-weighted estimate of habitat use within each home range. Company, Newport, Kentucky). Adult fox squirrels weighing > We only included fox squirrels with ≥ 30 relocations during

750 g (collar weight [19 g] was ≤ 3% body weight; Model SI-2C, our study period (March 2011–June 2012) in this analysis. Downloaded from Holohil Systems Ltd., Ontario, Canada) were radiocollared and Individual fox squirrels were treated as the sampling unit, and released at the capture location. We had 33 radiocollars avail- we considered all cover types simultaneously. Cover-type data able for deployment, and we trapped periodically from February were from the North Carolina Corporate Geographic Database 2011 to May 2012 to maintain 33 radiocollared fox squirrels (10-m resolution—Earth Satellite Corporation 1997). We used

throughout the study period. When a fox squirrel died, the collar 7 cover types based on dominant vegetation: southern yellow http://jmammal.oxfordjournals.org/ was retrieved and redeployed on another fox squirrel. All capture pine (primarily longleaf pine), bottomland hardwood forest, and processing methods met the specifications set forth by the managed herbaceous cover, mixed hardwood/conifers, mixed Institutional Animal Care and Use Committee at North Carolina shrubland, mixed upland hardwoods, and upland herbaceous State University (IACUC # 10-153-O) and followed guidelines (Table 1). We estimated fixed kernel density home ranges using of the American Society of Mammalogists (Sikes et al. 2011). Geospatial Modeling Environment (GME—Beyer 2012) and We relocated radiocollared fox squirrels once per day and at output UD grids with a 10 × 10-m cell size. We used the bivari- least 3 times per week using the homing technique at random ate plug-in option within GME to estimate the bandwidth for times between 0.5 h after sunrise and 0.5 h before sunset (White each fox squirrel’s kernel density estimate (Wand and Jones and Garrott 1990). If a squirrel was actively moving away from 1995; Gitzen et al. 2006). Use values were assigned based on us as we were approaching it, we stopped tracking and esti- the 95% UD, where the proportion of UD volume in each cover by James Harper on April 21, 2016 mated its original location using signal strength and direction. type represented an individual’s habitat use within the home In addition, we recorded whether a radiosignal was active or range (Millspaugh et al. 2006). We tested the null hypoth- inactive before homing to each squirrel. The majority (> 75%) esis (i.e., no selection) using multivariate analysis of variance of squirrels were inactive before we tracked to them, meaning (Wilks’ lambda). Rejection of the null hypothesis led to a series they were not actively moving away or had already frozen in of paired t-tests that ranked cover types from most to least place before we started tracking. Radiocollared fox squirrels selected (Aebischer et al. 1993). We used the adehabitatHS were monitored continually until death, radio failure, or they package within program R (R Development Core Team 2012) could no longer be tracked because they had moved into an to implement compositional analysis and to rank cover types artillery impact area. All fox squirrel relocations were recorded within the study area (Calenge 2006). using a handheld GPS (Rino120, Garmin International, Inc., We investigated habitat selection within each fox squirrel’s Olathe, Kansas). home range using resource utilization functions (RUFs— To assess the importance of hardwoods (especially oaks) Marzluff et al. 2004). We related UDs to habitat covari- on habitat selection by fox squirrels, we randomly selected a ates believed to influence habitat selection by fox squirrels

Table 1.—Descriptions of the 7 cover types used in analyses of resource selection by southeastern fox squirrels on Fort Bragg, North Carolina, 2011–2012.

Cover type Description Bottomland hardwoods Lowland areas with deciduous dominant woody vegetation ≥ 3 m in height and crown density ≥ 25%. Managed herbaceous cover Actively managed areas of herbaceous cover, including drop zones and artillery firing points. Mixed hardwoods/conifers Areas with ≥ 25% intermixture of deciduous and evergreen species. Hardwoods constitute a plurality of stocking, but pines account for 25–50% of the stocking. Mixed shrubland Areas with vegetation (evergreen and/or deciduous) dominated by shrubs and/or woody plants < 3 m in height. Mixed hardwoods Upland areas with deciduous dominant woody vegetation > 3 m in height and crown density ≥ 25%. Southern yellow pine Forested areas with 75% pine, including longleaf pine, loblolly slash pine, and/or pond pine. Upland herbaceous Unmanaged upland areas covered by herbaceous vegetation. 634 JOURNAL OF MAMMALOGY using multiple regression adjusted for spatial autocorrelation highly correlated with count of oaks (r > 0.75), so we removed (Marzluff et al. 2004). Habitat covariates included number of them from the analysis. We used the sign of parameter esti- large hardwood trees (≥ 26 cm DBH) per hectare, number of mates to indicate directionality of effect for independent vari- large pine trees (≥ 37 cm DBH) per hectare, number of grow- ables. Results were considered to be statistically significant ing-season burns in the previous 20 years, distance to near- at α = 0.05. We performed the analysis in JMP Pro 10 (SAS est riparian area (m), and distance to nearest road (m). Roads Institute Inc. 2012, Cary, North Carolina). included paved surfaces, unpaved surfaces, and firebreaks, and riparian areas were defined by the presence of permanent wet- land vegetation (Fort Bragg GIS Database). We only included Results fox squirrels with ≥ 30 relocations in the RUF analysis. Using From January 2011 to May 2012, we captured 76 (28 F, 47 M, the isopleth command in GME, we converted each UD to 99% 1 Unk) fox squirrels on Fort Bragg and equipped 52 (20 F, volume contour polygons, where contours represented 1–99 31 M, 1 Unk) with radiocollars. One squirrel (unknown sex) percentiles of use probabilities. We overlaid 30 × 30-m sam- was released before recording sex on the data form. Forty-eight pling grids centered on the habitat raster layers on each percent fox squirrels (22 F, 25 M, 1 Unk) had sufficient relocations to

volume polygon within ArcGIS10 (ESRI 2012). The sample include in the analysis of habitat selection. For fox squirrels Downloaded from tool was used within ArcGIS10 to extract relative use values included in our analyses, the number of relocations ranged and covariates associated with each point in the sampling grid. from 30 to 208 ( X = 116 ± 62), and tracking times varied from We acquired GIS layers from Fort Bragg personnel that con- 54 to 452 (X = 243 ± 137) days. tained all habitat covariates used in the analyses. Fox squirrels used cover types at Fort Bragg in a nonrandom

For the RUF analysis, we used the ruf package within pro- manner (Ʌ = 0.02, d.f. = 2, P = 0.001) when establishing home http://jmammal.oxfordjournals.org/ gram R (Handcock 2012). We used each fox squirrel’s band- ranges. We detected the following order of selection: southern width estimate from the bivariate plug-in as the starting point yellow pine > mixed hardwood/conifer > bottomland hardwood for estimating the range of spatial dependence and used 1.5 for > upland herbaceous > mixed shrubland > mixed hardwood the smoothing estimate within the ruf.fit function. We evaluated > managed herbaceous (Table 2). However, the differences the need for transformations by examining the residual plots between mixed hardwood/conifer and bottomland hardwood, from univariate RUFs for 5 randomly selected fox squirrels. and upland herbaceous, mixed shrubland, mixed hardwood, The response variable for all fox squirrels was log-transformed, and managed herbaceous were not significant. which normalized the response variable and residuals from the Distance to riparian area was the only significant predictor univariate RUFs. All covariates were used to create RUFs for of fox squirrel habitat use in the population-level RUF; use each fox squirrel and we averaged the standardized coefficients increased with decreasing distance to a riparian area (Table 3). by James Harper on April 21, 2016 to create a population-level RUF. We used the relative use value Predictor variables were not consistently correlated with use as the dependent variable and habitat covariates as the indepen- among individual fox squirrels, and predictor variables varied dent values in the multiple regression analysis. The magnitude in degree of importance among squirrels (Table 3). The logistic and sign of the standardized coefficients indicated the impor- regression model (X2 = 41.58, d.f. = 5, P ≤ 0.0001) indicated that tance of resources and the direction of use, respectively. hardwood DBH and largest pine DBH were negatively associ- We also investigated the relative importance of pine and ated with fox squirrel locations, whereas the count of oak trees hardwood trees in explaining fox squirrel relocations. We used was positively associated with fox squirrel locations (Table 4). logistic regression with fox squirrel relocations (set to 1) and random points (set to 0) as the binary response. We tested Discussion pine DBH, hardwood DBH, largest pine DBH, count of trees, and count of oak trees as independent variables. Originally, Fox squirrels selected home ranges composed of upland yellow we included oak DBH and largest oak DBH, but both were pine, a cover type that dominated the fire-maintained systems

Table 2.—Cover-type selection rankings for selection of home ranges by southeastern fox squirrels on Fort Bragg, North Carolina, 2011–2012. The sign of the t statistic is indicated with + or − signs; +++ and −−− represent significant deviation from zero.

Cover type Cover type Bottomland Managed Mixed hardwood/ Mixed shrubland Mixed hardwood Southern yellow Upland Rank hardwood herbaceous conifer pine herbaceous Bottomland hardwood • +++ − +++ +++ −−− +++ 3 Managed herbaceous −−− • −−− + +++ −−− − 7 Mixed hardwood/ + +++ • +++ +++ −−− +++ 2 conifer Mixed shrubland −−− − −−− • + −−− − 5 Mixed hardwood −−− −−− −−− − • −−− −−− 6 Southern yellow pine +++ +++ +++ +++ +++ • +++ 1 Upland herbaceous −−− + −−− + +++ −−− • 4 PRINCE ET AL.—SOUTHEASTERN FOX SQUIRREL RESOURCE SELECTION 635

Table 3.—Standardized resource utilization function coefficients for 48 radiocollared southeastern fox squirrels and the number of individual squirrels with significant positive (+) and negative (−) use associated with each habitat variable on Fort Bragg, North Carolina, 2011–2012.

Number of individuals Variable Standardized β 95% CI P ( β = 0) + − Distance to riparian area (m) −0.057 −0.108, −0.006 0.035 4 13 Distance to road (m) 0.032 −0.009, 0.073 0.135 9 6 Growing-season firesa −0.010 −0.028, 0.008 0.277 3 4 Large pine trees/ha 0.011 −0.014, 0.035 0.415 3 2 Large hardwood trees/ha 0.003 −0.025, 0.030 0.860 2 0 aNumber of fires in the last 20 years.

Table 4.—Parameters of the logistic regression model of habitat heterogeneously burned forests may help distribute resources selection by southeastern fox squirrels on Fort Bragg, North Carolina, across forest patches. In Florida, Kantola and Humphrey (1990)

2011–2012. observed Sherman’s fox squirrels using ecotones between pine Downloaded from stands and hardwood stands. Further, Abert’s squirrel (S. aberti), β SE X 2 P value once believed to be dependent on ponderosa pine (P. ponderosa) Intercept 1.0646 0.2421 19.34 < 0.0001 forests (that have changed in ecological structure at least in part Pine DBH 0.0001 0.0010 0.01 0.9072 Hardwood DBH −0.0052 0.0013 16.51 < 0.0001 due to fire suppression—seeAllen et al. 2002), showed adapt-

Largest pine DBH −0.0514 0.0118 18.72 < 0.0001 ability to other tree species and forest types, indicating selection http://jmammal.oxfordjournals.org/ Count of trees −0.0112 0.0084 1.78 0.1823 for structural components of forests (Edelman and Koprowski Count of oaks 0.0488 0.0158 9.59 0.0020 2005). Indeed, in our study, fox squirrels depended on both pine and hardwood components for dietary and cover requirements, which indicate that burn heterogeneity on the landscape may in which fox squirrels evolved (Weigl et al. 1989; Perkins and improve habitat quality for fox squirrels. Conner 2004; Perkins et al. 2008). During summer months, Similar to other reports of oaks being important to fox squir- fox squirrels feed heavily on seeds obtained from longleaf rel resource selection (Weigl et al. 1989; Conner and Godbois pine cones, commonly consuming 20–30 cones per day (Steele 2003), squirrel relocations in our study were positively asso- 1988; Weigl et al. 1989; Steele and Weigl 1993). Fox squir- ciated with increasing numbers of reproductively mature oak rels are influenced by understory vegetation height and have stems. However, we detected a negative association between by James Harper on April 21, 2016 been observed to feed on the ground close to 70% of the time fox squirrel relocations and hardwood DBH and largest pine (Ditgen et al. 2007; Eisenberg et al. 2011). However, at finer DBH, which seems counterintuitive. We believe this was an scales of habitat selection, fox squirrels consistently selected artifact of our sampling, as most of the largest DBH hardwoods areas with a hardwood component, likely because of the food and pines were located in riparian areas at Fort Bragg which and cover resources they provided. Within longleaf pine stands, contain low densities of reproductively mature oaks (Lashley fox squirrels concentrated their activities near riparian areas, et al. 2014). Because fox squirrel relocations were often in which likely supported less fire-tolerant hardwood species that ecotones, paired random points commonly fell well inside the produce seasonally important cover and food resources (Hoctor riparian area and much closer to the large non-oak hardwoods et al. 2006). During our study, fox squirrels selected areas with (e.g., blackgum [Nyssa sylvatica], tulip poplar [Liriodendron more oak trees during all seasons of the year, highlighting the tulipifera]) and pines (e.g., loblolly) that are common in more importance of oaks for food, cover, and nesting sites, which mesic conditions. Thus, squirrel selection for sites with the is similar to other studies (Hilliard 1979; Kantola 1986; Weigl most oak trees, but not for the largest DBH hardwoods, sup- et al. 1989; Powers 1993). Perkins et al. (2008) suggested that ports the well-established importance of oaks to fox squirrels. optimal fox squirrel habitat contains 88.2% mature pine cover Chamberlain et al. (1999) found that fox squirrel abundance and 11.8% hardwood cover; they speculated that the upland increased with decreasing stand basal area, but only when hardwood component could be lower if pine stands were adja- percentage of hardwoods increased. Though we did not mea- cent to streams where hardwoods were more prevalent. At sure percentage of hardwood (or oak) in the stands, fire-main- Fort Bragg, hardwoods were removed mechanically in upland tained longleaf pine forests generally have low basal area in stands, potentially restricting mature mast-producing oaks to the uplands, which may help explain why fox squirrels at Fort riparian areas that did not burn or along firebreaks that provided Bragg showed selection for patches of oaks in uplands, but also a fire shadow (Lashley et al. 2014). concentrated use in close proximity to riparian areas. Fox squirrels in our study were similar to other tree squir- Because oaks and other hardwoods provide important rel species in their dependence on habitat heterogeneity. For food and cover resources, management actions (i.e., frequent example, Douman (2010) showed that Mexican fox squirrels high-intensity fires, chemical treatments) that limit the abun- (S. nayaritensis) used low-severity burned forests more heav- dance and distribution of mature hardwoods could decrease ily than those that burned at higher severities, but suggested fox squirrel survival and reproductive success (Weigl et al. 636 JOURNAL OF MAMMALOGY

1989; Conner and Godbois 2003). Nesting sites can include Baumgartner, L. L. 1940. Trapping, handling, and marking fox natural cavities and leaf nests and are critical to fox squirrels squirrels. Journal of Wildlife Management 4:444–450. for a variety of reasons, including protection against preda- Beyer, H. L. 2012. Geospatial modeling environment. www.spatiale- tors and shelter from poor weather conditions (Baumgartner cology.com/gme. Accessed 20 October 2012. 1939; Moore 1957; Nixon et al. 1984). In North Carolina, fox Boyer, W. D. 1990. Growing-season burns for control of hardwoods in longleaf pine stands. USDA Forest Service, Southern squirrels used artificial cavities (nest boxes) more often in Forest Experiment Station, Research Paper SO-256, New Orleans, rainy or cold weather and during periods of low food supply Louisiana. (Weigl et al. 1989). Weigl et al. (1989) hypothesized that a Brennan, L. A. 1991. How can we reverse the northern bobwhite lack of cavities for rearing young in upland areas was limit- population decline? Wildlife Society Bulletin 19:544–555. ing fox squirrel populations in North Carolina. Evaluating Brennan, L. A., et al. 1998. Whither wildlife without fire? the effects of frequent prescribed fires during the growing Transactions of the North American Wildlife and Natural Resources season on fox squirrel reproductive success could help deter- Conference 63:402–414. mine whether relegation of hardwood stems to riparian areas Brockway, D. G., and C. E. Lewis. 1997. Long-term effects of dor- provides the food and cover needed to sustain fox squirrel mant-season prescribed fire on plant community diversity, structure and productivity in a longleaf pine wiregrass ecosystem. populations. Downloaded from Frequent fires in upland pine stands likely benefited fox Forest Ecology and Management 96:167–183. Buckner, E. 1989. Evolution of forest types in the southeast. Pp. squirrels by preventing the succession of open, pine-dominated 27–33 in Proceedings of pine-hardwood mixtures: a symposium on uplands to high-density hardwood stands favored by gray management and ecology of the type (T. A. Waldrop, ed.). USDA squirrels. When developing a prescribed-fire program, resource Forest Service, Southeastern Forest Experiment Station, General managers should strive to leave burns patchy, allowing for the Technical Report SE-58, Asheville, North Carolina. http://jmammal.oxfordjournals.org/ regeneration and maturation of oaks and other hardwoods in Calenge, C. 2006. The package adehabitat for the R software: a tool areas of low topography and mesic areas adjacent to streams for the analysis of space and habitat use by animals. Ecological (i.e., in fire shadows). A frequent growing-season fire regime Modeling 197:516–519. that suppresses hardwoods completely within upland pine Cantrell, M. A., J. J. Britcher, and E. L. Hoffman. 1995. Red- stands should be balanced with low-intensity fires in riparian cockaded woodpecker management initiatives at Fort Bragg areas that shelter less fire-tolerant hardwood species. If riparian Military Installation. Pp. 89–97 in Red-cockaded woodpecker: areas with canopy-level hardwoods are not widely distributed recovery, ecology, and management (D. L. Kulhavy, R. G. Hooper, and R. Costa, eds.). Center for Applied Studies in Forestry, College across a site, managers striving to maintain structural condi- of Forestry, Stephen F. Austin State University, Nacogdoches, tions that benefit fox squirrels should leave mature hardwoods Texas. by James Harper on April 21, 2016 scattered throughout fire-maintained upland pine stands, rather Chamberlain, M. J., J. M. Ross, and B. D. Leopold. 1999. Influence than completely eliminating them through mechanical removal of forest management and microhabitat conditions on abundance or frequent growing-season fires. of southern fox and gray squirrels. Proceedings of the Annual Conference of Southeastern Association of Fish and Wildlife Acknowledgments Agencies 53:402–414. Christensen, N. L. 1981. Fire regimes in southeastern eco- Funding was provided by the United States Department of systems. Pp. 112–135 in Proceedings of the conference fire Defense, Fort Bragg Wildlife Branch, and North Carolina State regimes and ecosystem properties (H. A. Mooney, T. M. University’s Fisheries, Wildlife, and Conservation Biology Bonnicksen, N. L. Christensen, J. E. Lotan, and W. A. Reiners, Program. We thank J. Jones and A. Shultz (Fort Bragg Wildlife eds.). USDA Forest Service, General Technical Report WO-26, Branch) for providing technical and logistical support. We Washington, D.C. thank C. Farr, M. Frisicano, R. Davis, B. Sherrill, M. Broadway, Conner, L. M., and I. A. Godbois. 2003. Habitat associated with day- M. Bennett, and J. Nevins for assistance in the field. We thank time refugia of fox squirrels in a longleaf pine forest. American K. Aubry and several anonymous reviewers for comments on Midland Naturalist 150:123–129. Conner, L. M., J. L. Landers, and W. K. Michener. 1999. Fox squir- the manuscript. rel and gray squirrel associations within minimally disturbed longleaf pine forests. Proceedings of the Annual Conference Literature Cited of Southeastern Association of Fish and Wildlife Agencies Aebischer, N. J., P. A. Robertson, and R. E. Kenward. 1993. 53:364–374. Compositional analysis of habitat use from animal radio‐tracking Ditgen, R. S., J. D. Shepherd, and S. R. Humphrey. 2007. Big Cypress data. Ecology 74:1313–1325. fox squirrel (Sciurus niger avicennia) diet, activity, and habitat use Allen, D. L. 1943. Michigan fox squirrel management. Michigan on a golf course in southwest Florida. American Midland Naturalist Department of Conservation, Game Division Publication 100, 158:403–414. Lansing, Michigan. Douman, S. L. 2010. Return of historical fire: impacts of burn severity Allen, C. D., et al. 2002. Ecological restoration of southwest- and heterogeneity on Mexican fox squirrels. M.S. thesis, University ern ponderosa pine ecosystems: a broad perspective. Ecological of Arizona, Tucson. Applications 12:1418–1433. Earth Satellite Corporation. 1997. Comprehensive land cover map- Baumgartner, L. L. 1939. Foods of the fox squirrel in Ohio. ping for the state of North Carolina: final report. North Carolina Transactions of the North American Wildlife and Natural Resources Center for Geographic Information and Analysis (NCCGIA) dis- Conference 4:579–584. tributed the data, Rockville, Maryland. PRINCE ET AL.—SOUTHEASTERN FOX SQUIRREL RESOURCE SELECTION 637

Edelman, A. J., and J. L. Koprowski. 2005. Selection of drey sites Kantola, A. T., and S. R. Humphrey. 1990. Habitat use by Sherman’s by Abert’s squirrels in an introduced population. Journal of fox squirrel (Sciurus niger shermani) in Florida. Journal of Mammalogy 86:1220–1226. Mammalogy 71:411–419. Eisenberg, D. A., R. F. Noss, J. M. Waterman, and M. B. Main. Koprowski, J. L. 2002. Handling tree squirrels with a safe and effi- 2011. Distribution and habitat use of the Big Cypress fox squirrel cient restraint. Wildlife Society Bulletin 30:101–103. (Sciurus niger avicennia). Southeastern Naturalist 10:75–84. Kush, J. S., R. S. Meldahl, and A. Chadwick. 2004. A restoration Engstrom, R. T., D. B. McNair, L. A. Brennan, C. L. Hardy, and success: longleaf pine seedlings established in a fire-suppressed, L. W. Burger. 1996. Influence of dormant versus lightning sea- old-growth stand. Ecological Restoration 22:6–10. son prescribed fire on birds in longleaf pine forests: experimental Landers, J. L., N. A. Byrd, and R. Komarek. 1990. A holistic design and preliminary results. Transactions of the North American approach to managing longleaf pine communities. Proceedings of Wildlife and Natural Resources Conference 61:200–207. the symposium on the management of longleaf pine. USDA Forest ESRI. 2012. ArcGIS. Ver. 10.0. Environmental System Research Service, Southern Forest Experiment Station, General Technical Institute, Inc., Redlands, California. Report SO-75, New Orleans, Louisiana. Fill, J. M., S. M. Welch, J. L. Waldron, and T. A. Mousseau. 2012. Landers, J. L., D. H. Van Lear, and W. D. Boyer. 1995. The longleaf The reproductive response of an endemic bunchgrass indicates his- pine forests of the Southeast: requiem or renaissance? Journal of torical timing of a keystone process. Ecosphere 3:1–12. Forestry 9:39–44. Downloaded from Frost, C. C. 1993. Four centuries of changing landscape patterns in Lashley, M. A., et al. 2014. Subtle effects of a managed fire regime: the longleaf pine ecosystem. Proceedings of the Tall Timbers Fire a case study in the longleaf pine ecosystem. Ecological Indicators Ecology Conference 18:17–44. 38:212–217. Frost, C. C. 1998. Presettlement fire frequency regimes of the United Loeb, S. C., and N. D. Moncrief. 1993. The biology of fox squirrels States: a first approximation. Proceedings of the Tall Timbers Fire (Sciurus niger) in the Southeast: a review. Pp. 1–19 in Proceedings

Ecology Conference 20:70–81. of the second symposium on southeastern fox squirrels, Sciurus http://jmammal.oxfordjournals.org/ Frost, C. C. 2006. History and future of the longleaf pine ecosystem. niger (J. W. Edwards and P. A. Tappe, eds.). Virginia Museum of Pp. 9–48 in Longleaf pine ecosystems: ecology, management, and Natural History, Special Publication 1, Martinsville. restoration. (S. Jose, E. Jokela, and D. Miller, eds.). Springer, New MacCleery, D. W. 1993. American forests--a history of resiliency York. and recovery. Revised ed. USDA Forest Service and the Forest Garren, K. H. 1943. Effects of fire on vegetation of the southeastern History Society, Durham, North Carolina. United States. Botanical Review 9:617–654. Marzluff, J. M., J. J. Millspaugh, P. Hurvitz, and M. S. Handcock. Gitzen, R. A., J. J. Millspaugh, and B. J. Kernohan. 2006. Bandwidth 2004. Relating resources to a probabilistic measure of space use: selection for fixed kernel analysis of animal range use. Journal of forest fragments and Steller’s jays. Ecology 85:1411–1427. Wildlife Management 70:1334–1344. Means, D. B. 1996. Longleaf pine forest, going, going . . . Pp. 210– Greenberg, C. H., and R. W. Simons. 1999. Age, composition, and 229 in Eastern old-growth forest: prospects for rediscovery and by James Harper on April 21, 2016 stand structure of old-growth oak sites in the Florida high pine recovery (M. B. Davis, ed.). Island Press, Washington, D.C. landscape: implications for ecosystem management and restora- Millspaugh, J., R. Nielson, and L. McDonald. 2006. Analysis tion. Natural Areas Journal 19:30–40. of resource selection using utilization distributions. Journal of Ha, J. C. 1983. Food supply and home range in the fox squirrel Wildlife Management 70:384–395. (S. niger). M.S. thesis, Wake Forest University, Winston-Salem, Moore, J. C. 1957. The natural history of the Sherman’s fox squir- North Carolina. rel (Sciurus niger shermani). Bulletin of the American Museum of Handcock, M. S. 2012. Estimates of the resource utilization func- Natural History 113:1–71. tion. Version 1.5-2. http://www.csde.washington.edu/~handcock/ Nixon, C. M., S. P. Havera, and L. P. Hansen. 1984. Effects of nest ruf. Accessed 12 January 2013. boxes on fox squirrel demography, condition, and shelter use. Heyward, F. 1939. The relation of fire to stand composition of long- American Midland Naturalist 112:157–171. leaf pine forests. Ecology 20:287–304. Nowacki, G. J., and M. D. Abrams. 2008. The demise of fire and Hiers, J. K., et al. 2014. Ecological value of retaining pyrophytic “mesophication” of forests in the eastern United States. BioScience oaks in longleaf pine ecosystems. Journal of Wildlife Management 58:123–138. 78:383–393. Outcalt, K. W., and R. M. Sheffield. 1996. The longleaf pine for- Hilliard, T. H. 1979. Radio-telemetry of fox squirrel in the Georgia est: trends and current conditions. USDA Forest Service, Southern Coastal Plain. M.S. thesis, University of Georgia, Athens. Research Station, Resource Bulletin SRS-9, Asheville, North Hoctor, T. S., R. F. Noss, L. D. Harris, and K. A. Whitney. 2006. Carolina. Spatial ecology and restoration of the longleaf pine ecosystem. Pp. Perkins, M. W., and L. M. Conner. 2004. Habitat use of fox squir- 377–402 in Longleaf pine ecosystems: ecology, management, and rels in southwestern Georgia. Journal of Wildlife Management restoration (S. Jose, E. Jokela, and D. Miller, eds.). Springer, New 68:509–513. York. Perkins, M. W., L. M. Conner, and M. B. Howze. 2008. The impor- Hudson, C. M. 1982. The southeastern Indians. University of tance of hardwood trees in the longleaf pine forest ecosystem Tennessee Press, Knoxville. for Sherman’s fox squirrels. Forest Ecology and Management James, F. C., and H. H. Shugart, Jr. 1970. A quantitative method of 255:1618–1625. habitat description. Audubon Field Notes 24:727–736. Powers, J. S. 1993. Fox squirrel home range and habitat use in Kantola, A. T. 1986. Fox squirrel home range and mast crops in the Southeastern Coastal Plain. M.S. thesis, Auburn University, Florida. M.S. thesis, University of Florida, Gainesville. Auburn, Alabama. Kantola, A. T. 1992. Sherman’s fox squirrel. Pp. 234–241 in Rare Provencher, L., et al. 2001. Longleaf pine and oak responses to and endangered biota of Florida, vol. 1, mammals (S. R. Humphrey, hardwood reduction techniques in fire-suppressed sandhills in ed.). University Press of Florida, Gainesville. northwest Florida. Forest Ecology and Management 148:1–15. 638 JOURNAL OF MAMMALOGY

Pyne, S. J. 1982. Fire in America, a cultural history of wildland and Steele, M., and P. Weigl. 1993. The ecological significance of body rural fire. University of Washington Press, Seattle. size in fox squirrels (Sciurus niger) and gray squirrels (Sciurus R Development Core Team. 2012. R: a language and environment carolinensis). Pp. 57–69 in Proceedings of the second symposium for statistical computing. R Foundation for Statistical Computing, on southeastern fox squirrels, Sciurus niger (N. D. Moncrief, J. Vienna, Austria. www.R-project.org/. Accessed 10 August 2012. W. Edwards, and P. A. Tappe, eds.). Virginia Museum of Natural Rebertus, A. J., G. B. Williamson, and W. J. Platt. 1993. Impacts History, Special Publication 1, Martinsville. of temporal variation in fire regime on savanna oaks and pines. Stoddard, H. L., Sr. 1962. Use of fire in pine forests and game lands Proceedings of the Tall Timbers Fire Ecology Conference of the deep southeast. Proceedings of the Tall Timbers Fire Ecology 18:215–225. Conference 1:31–42. Robbins, L. E., and R. L. Myers. 1992. Seasonal effects of prescribed Varner, J. M., D. R. Gordon, F. E. Putz, and J. K. Hiers. 2005. Novel burning in Florida: a review. Tall Timbers Research Station, fire effects in south-eastern pine forests: smoldering fire and over- Miscellaneous Publication 8, Tallahassee, Florida. story pine mortality. Restoration Ecology 13:539–544. SAS Institute Inc. 2012. JMP Pro 10. SAS Institute Inc., Cary, North Wand, M. P., and M. C. Jones. 1995. Kernel smoothing. Chapman Carolina. and Hall, London, United Kingdom. Scott, J. A. 2011. Empirical forecast models for guiding decision- Weigl, P. D., M. A. Steele, L. J. Sherman, J. C. Ha, and T. making in complex landscapes with competing wildlife manage- L. Sharpe. 1989. The ecology of the fox squirrel (Sciurus ment goals. Ph.D. dissertation, University of Georgia, Athens. niger) in North Carolina: implications for survival in the Downloaded from Sikes, R. S., W. L. Gannon, and The Animal Care and Use Committee Southeast. Bulletin of Tall Timbers Research Station Number of the American Society of Mammalogists. 2011. Guidelines of 24, Tallahassee, Florida, the American Society of Mammalogists for the use of wild mam- Whitaker, J. O., and W. J. Hamilton. 1998. Mammals of the eastern mals in research. Journal of Mammalogy 92:235–253. United States. Cornell University Press, Ithaca, New York.

Sorrie, B. A., J. B. Gray, and P. J. Crutchfield. 2006. The vascular White, G. C., and R. A. Garrott. 1990. Analysis of wildlife radio- http://jmammal.oxfordjournals.org/ flora of the longleaf pine ecosystem of Fort Bragg and Weymouth tracking data. Academic Press, San Diego, California. Woods, North Carolina. Southern Appalachian Botanical Society 71:129–161. Submitted 14 March 2015. Accepted 15 December 2015 Steele, M. A. 1988. Patch use and foraging ecology of the fox squirrel: tests of theoretical predictions. Ph.D. dissertation, Wake Forest Associate Editor was Keith B. Aubry. University, Winston-Salem, North Carolina. by James Harper on April 21, 2016